Our IRIs for import-profiles (AOB for today - ideally)

Dear all,

we need in SPARQL IRIs to refer to entailment regimes as defined by
OWL and RDF, e.g.
simple entailment, RDF entailment, RDFS entailment, various OWL
fragments, etc...

We (kind of) have IRIs for those already in RIF, by the import-profile
IRIs. It looks like it makes sense to unify these and arrive at
unified IRIS referring to these entailment regimes reusable for
both groups (or other future purposes).

So, can we still change the IRIs for import profiles to some generic
IRIs?

At the moment, we use:

http://www.w3.org/2007/rif-import-profile#Simple
http://www.w3.org/2007/rif-import-profile#RDF
http://www.w3.org/2007/rif-import-profile#RDFS
http://www.w3.org/2007/rif-import-profile#D
http://www.w3.org/2007/rif-import-profile#OWL-DL
http://www.w3.org/2007/rif-import-profile#OWL-Full

Ideally, we'd have some generic IRIs here, referring to the resp.
*Entailment regime* rather than to an *import profile*.
The reason why I ask this is that generic IRIs for the different
Entailment regimes would be reusable in other contexts
by other groups as well. In SPARQL we currently discuss IRIs for the
entailment regimes in a similar context,
for SPARQL/RDFS SPARQL/OWL, etc...

In my opinion, it would be awkward to come up there with own IRIs, e.g.

http://www.w3.org/2009/sparql-entailment#Simple
http://www.w3.org/2009/sparql-entailment#RDF
http://www.w3.org/2009/sparql-entailment#RDFS
http://www.w3.org/2009/sparql-entailment#D
http://www.w3.org/2009/sparql-entailment#OWL-DL
http://www.w3.org/2009/sparql-entailment#OWL-Full
...

but it would be much preferable to refer to unified, generic IRIs for
the resp. entailment
regimes in both cases  (and for the good of the SW languages family to
live happily ever after), e.g.

http://www.w3.org/2009/entailment-regime#Simple
http://www.w3.org/2009/entailment-regime#RDF
http://www.w3.org/2009/entailment-regime#RDFS
http://www.w3.org/2009/entailment-regime#D
http://www.w3.org/2009/entailment-regime#OWL-DL
http://www.w3.org/2009/entailment-regime#OWL-Full


For the choice of such generic IRIs there are obviously several
alternatives concerning the namespace

* Option1: use a new namespace like above, i.e.

http://www.w3.org/2009/entailment-regime#
or
http://www.w3.org/2009/rdf-owl-entailment-regime#

+ neutral, reusable
- yet another new namespace

* Option2: use the "normal" RIF namespace, i.e., just remove "import-
profile"

     http://www.w3.org/2007/rif#

+ looks like a clean solution
-  might raise concerns of being "RIF specific"

* Option3:  to reuse the RDFS and OWL namespaces here, e.g.

http://www.w3.org/2002/07/owl#OWL_DL
http://www.w3.org/2002/07/owl#OWL_2_DL
http://www.w3.org/2002/07/owl#OWL_2_EL
http://www.w3.org/2002/07/owl#OWL_2_QL
http://www.w3.org/2002/07/owl#OWL_2_RL
http://www.w3.org/2002/07/owl#OWL_Full
http://www.w3.org/2002/07/owl#OWL_2_Full
http://www.w3.org/2000/01/rdf-schema#Simple
http://www.w3.org/2000/01/rdf-schema#RDF
http://www.w3.org/2000/01/rdf-schema#RDFS
http://www.w3.org/2000/01/rdf-schema#D

+ looks like the cleanest option to me
- might raise eyebrows of extending/reusing other namespaces, we'd
need approval from OWL at least (I'd be hopeful we could get that, but
not sure what to do re:RDFS,
    one variant/workaround this would be to use the OWL namespace for
all of these, which I wouldn't mind to, cf. Option 3b)

Option 3b: like Option3, but uniformly use the OWL namespace, i.e.
...
http://www.w3.org/2002/07/owl#Simple
http://www.w3.org/2002/07/owl#RDF
http://www.w3.org/2002/07/owl#RDFS
http://www.w3.org/2002/07/owl#D

+ looks like a clean solution, we'd only need to ask OWL whether
that's ok for them
    I actually favor that option at the moment


I would like to raise this as an agenda item in the next RIF telco.

best,
Axel

Received on Tuesday, 13 October 2009 14:57:52 UTC